Computer Vision Group Project

Overview

Built augmented reality system with dynamic object tracking using OpenCV and Intel RealSense for spatial data, achieving <100ms inference latency.

Problem

AR applications require real-time object tracking and pose estimation in challenging environments with low latency for natural user experience.

Solution

Developed end-to-end CV pipeline integrating sensor data (Intel RealSense) with TensorFlow MoveNet for pose estimation. Implemented OpenCV-based tracking algorithms for dynamic object detection and AR visualization. 3-person team collaboration.

Impact

• <100ms latency: Real-time pose estimation on standard hardware
• Robust tracking: Dynamic object detection in challenging conditions
• Multimodal integration: RealSense depth + RGB processing
• Open source: Published collaborative team implementation